Phillips, Wellman, and Spelke (2002) provided experimental evidence indicating that by the age of 12 months infants use information about an adult's gaze-direction and emotional expression to predict action. We investigate the generality of this ability, using Phillips et al.'s paradigm across different referential gestures. If infants have a rich concept of people as intentional agents, they should encode the person-goal relation and predict action from a variety of referential gestures (such as looking and pointing), as well as from (goal-directed) reaching and grasping actions. Results of Studies 1 and 2 (within-subjects design) and Study 3 (between-subjects design) converge in indicating that Phillips et al.'s basic finding generalizes to reach-and-gaze as cue condition. In Study 3b generalization was demonstrated for point-and-gaze as cue condition. Whereas these findings suggest a general understanding of referential intent at 12 months, a control experiment (Study 4) showed that infants' action expectations could be reversed following a psychologically inconsistent habituation phase (actor points at Object A and subsequently picks up Object B) in the point-and-gaze condition, although no such reversal was obtained in the reach-and-gaze condition. These findings indicate that 1-year-old infants have well-established expectations about action goals signaled by an actor's gaze-direction and reaching and grasping movements but that an understanding of pointing as a cue to action intentions may develop later and follow a different route of acquisition. (Contains 6 figures and 4 footnotes.)